Coronary Angiogram Multi-View Matching Using Relational Networks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods struggle to effectively address the challenges of determining correspondences between coronary angiogram images, particularly in the context of determining correspondences between coronary angiogram images, particularly in the context of determining correspondences between coronary angiogram images, with coronary angiogram images, specifically in the field of determining correspondences between angiogram view matching, particularly in the field of matching angiogram images.

Innovation Solution

A machine learning-based approach is employed to automatically determine corresponding locations of points of interest across multiple coronary angiogram images using a machine learning-based relational network, such as a transformer neural network, to relate features and identify the locations of anatomical landmarks or abnormalities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional methods are used to determine correspondences between coronary angiogram images, then user intervention is required which increases time consumption and operational complexity, but the system can handle the complex vessel structures and motion-related challenges with manual guidance

Engineering Contradiction:
Improveprocessing speedVSAvoidautomation level
Core Design Contradiction:
ProductivityVSExtent of automation

Solution Approach 1:

The system performs multi-view matching automatically without requiring user intervention. The machine learning model extracts features from angiogram images and determines correspondences autonomously, allowing the system to serve itself by eliminating the need for manual point selection and correspondence determination by clinicians.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical operations (user clicking, dragging, and selecting points) with an automated machine learning system. The neural network model processes images and automatically identifies corresponding locations, substituting the mechanical interaction between user and interface with an automated computational system.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If manual methods are used for multi-view matching, then user input can guide the process through complex vessel structures and motion artifacts, but the process becomes time-consuming and less efficient

Engineering Contradiction:
Improvematching accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary feature extraction from multiple angiogram images before performing the actual matching operation. By pre-processing the images and extracting relevant features in advance, the system prepares the data structure needed for rapid and accurate correspondence determination, reducing the time required for the actual matching task.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The machine learning model creates feature representations (copies) of the complex vessel structures and motion patterns from the input images. These feature copies capture the essential characteristics needed for matching without requiring the system to re-analyze the raw complex images during the matching process, thereby improving both accuracy and efficiency.

Inventive Principle:
Principle #26Copying

3Extent of automation

If automated machine learning methods are implemented for multi-view matching, then processing efficiency and automation are improved, but the system must accurately handle complex coronary vessel structures and motion-related challenges without user guidance

Engineering Contradiction:
Improveautomation levelVSAvoidsystem complexity
Core Design Contradiction:
Extent of automationVSDevice complexity

Solution Approach 1:

The patent segments the complex task of multi-view matching into distinct processing stages: feature extraction from individual images, relational network processing to establish correspondences, and final location determination. This segmentation allows each component to be optimized independently and simplifies the overall system architecture by breaking down the complex automated process into manageable modules.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces feature representations as intermediary data structures between the input angiogram images and the final correspondence results. These features act as mediators that capture the essential information from complex images with motion artifacts and vessel structures, transforming the complex recognition problem into a simpler matching problem in feature space.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4246457B1Multi-view matching across coronary angiogram images
Publication Date: 2026.04.29 SIEMENS HEALTHINEERS AG
  • EP4246457B1 patent drawingFigure 1
  • EP4246457B1 patent drawingFigure 2
  • EP4246457B1 patent drawingFigure 3

AI summary

Systems and methods for determining corresponding locations of points of interest in a plurality of input medical images are provided. A plurality of input medical images comprising a first input medical image and one or more additional input medical images is received. The first input medical image identifies a location of a point of interest. A set of features is extracted from each of the plurality of input medical images. Features between each of the sets of features are related using a machine learning based relational network. A location of the point of interest in each of the one or more additional input medical images that corresponds to the location of the point of interest in the first input medical image is identified based on the related features. The location of the point of interest in each of the one or more additional input medical images is output.